[MITgcm-support] global 4deg verification

[Martin Losch]

I may add, that you, if you are using anything later than release1, you 
could use a 2day  tracer timestep for even faster timestep, that is, 
for asynchronous (tracer accelerated) integration use deltaTmom = 1200, 
deltaTtracer=deltaTfreesurf=deltatclock=172800, and for synchronous 
timestepping use 3600 for all (deltaT=3600 works even for a two degree 
configuation).
For "true" equilibrium, run in asynchronous mode until the T/S-drift is 
negligible and then run in synchronous mode for another 30years to 
recover the seasonal cycle.

Bryan was the first to introduce the tracer acceleration:
Bryan, K., and L.J. Lewis, 1979: A water mass model of the world ocean. 
J. Geophys. Res., 84(C5), 2503-2517.
Bryan, K., 1984: Accelerating the convergence to equilibrium of 
ocean-climate models. J. Phys. Oceanogr., 14, 666-673.

BTW, Jean-Michel has implemented deep tracer acceleration 
(dTtracerLev(1:Nr) replaces  deltaTtracer), but so far there isn't any 
example experiment that you can start from, as far as I know.

Martin

On Jan 22, 2005, at 10:27 PM, samar khatiwala wrote:

> Chris, thanks for the reply.
>
> So just to be clear, I should:
>   1) run model with deltaTmom = 1200.0, deltaTtracer= 43200.0,
>      until close to equilibrium
>   2) run syncronously with deltaTmom=deltaTtracer=3600.0, until
>      equililibrium is reached
>
> You mention that there are different stability limits for sync and 
> async
> integrations. Does that mean, an asyncronous run with deltaTmom=3600
> and deltaTtracer= 43200 will be unstable? I know this is a black art,
> but it does seem a bit curious. Can you please point me to references
> to this issue?
>
> Thanks, Samar
>
> On Sat, 22 Jan 2005, chris hill wrote:
>
>> Hi Samar,
>>
>>   For a seasonal experiment you should change time steps to be the 
>> same.
>>   A 3600s timestep should be possible for the seasonal run. The 1200 
>> sec
>> is used to allow a longer async ratio, but for a synchronous run the
>> stability limits are different so a one hour dt should work.
>>
>> Chris
>> samar khatiwala wrote:
>>> Hello
>>>
>>> I have a query about verification/global_ocean.90x40x15
>>>
>>> As shipped, the momentum and tracer time steps in this experiment 
>>> are set
>>> to different values in 'data':
>>>
>>>  deltaTmom = 1200.0,
>>>  deltaTtracer= 43200.0,
>>>  deltaTClock = 43200.0,
>>>
>>> Is this because the default is asynchronous time stepping for more 
>>> rapid
>>> spin up? Once the model spins up to equilibrium, do I set 
>>> deltaTtracer,
>>> deltaTClock, deltaTmom equal to each other? I have read (Danabasoglu 
>>> et
>>> al) that in a seasonally forced model, the final spin uip must be 
>>> done
>>> with synchronous time stepping. Can I set all these deltaT's to 
>>> 43200.0
>>> sec without the model blowing up? Or does the CFL limit the time 
>>> step to
>>> 1200.0 sec.
>>>
>>> Thanks, Samar
>>> _______________________________________________
>>> MITgcm-support mailing list
>>> MITgcm-support at mitgcm.org
>>> http://dev.mitgcm.org/mailman/listinfo/mitgcm-support
>>>
>>
>> _______________________________________________
>> MITgcm-support mailing list
>> MITgcm-support at mitgcm.org
>> http://dev.mitgcm.org/mailman/listinfo/mitgcm-support
>>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://dev.mitgcm.org/mailman/listinfo/mitgcm-support